The Effects of Sanction Intensity on Criminal Conduct - JDAI Helpdesk
The Effects of Sanction Intensity on Criminal Conduct - JDAI Helpdesk
The Effects of Sanction Intensity on Criminal Conduct - JDAI Helpdesk
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
not the other. For example, the results <str<strong>on</strong>g>of</str<strong>on</strong>g> a set <str<strong>on</strong>g>of</str<strong>on</strong>g> experiments carried out in the United<br />
Kingdom to test the effect <str<strong>on</strong>g>of</str<strong>on</strong>g> face-to-face restorative justice <strong>on</strong> recidivism showed no<br />
change in the number <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders who participated in crime after the program compared<br />
to before. However, those <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders committed 27 per cent fewer crimes <strong>on</strong> average than<br />
they did prior to the program (Shapland et al., 2008). Looking at participati<strong>on</strong> al<strong>on</strong>e, <strong>on</strong>e<br />
might c<strong>on</strong>clude that restorative justice was no more effective than regular court<br />
processing.<br />
However, society still benefits if fewer crimes are committed overall.<br />
C<strong>on</strong>versely, the success <str<strong>on</strong>g>of</str<strong>on</strong>g> the low-intensity supervisi<strong>on</strong> strategy might be doubtful if the<br />
lack <str<strong>on</strong>g>of</str<strong>on</strong>g> difference between groups in the proporti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fenders participating in crime<br />
masked an increase in the frequency <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fending for the LIS group compared to those<br />
receiving SAU. Measures <str<strong>on</strong>g>of</str<strong>on</strong>g> participati<strong>on</strong> and frequency also produce different policyrelevant<br />
estimates <str<strong>on</strong>g>of</str<strong>on</strong>g> the treatment effect. In the present experiment, the participati<strong>on</strong><br />
measure gives the more accurate effect <str<strong>on</strong>g>of</str<strong>on</strong>g> assignment to LIS versus SAU, since LIS<br />
<str<strong>on</strong>g>of</str<strong>on</strong>g>fenders are returned to SAU after their first new <str<strong>on</strong>g>of</str<strong>on</strong>g>fense. However, if we c<strong>on</strong>tinue to<br />
follow experimental participants after they complete or fail LIS, the number <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fenses<br />
they commit in the l<strong>on</strong>ger term <str<strong>on</strong>g>of</str<strong>on</strong>g>fers an indicati<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> whether spending any amount <str<strong>on</strong>g>of</str<strong>on</strong>g><br />
time in LIS has a deterrent effect <strong>on</strong> subsequent criminal behavior.<br />
We assess the effect <str<strong>on</strong>g>of</str<strong>on</strong>g> the experimental treatment <strong>on</strong> participati<strong>on</strong> and frequency<br />
using regressi<strong>on</strong> models designed for binary and count data. We c<strong>on</strong>struct a binary logit<br />
model for participati<strong>on</strong>. Frequency <str<strong>on</strong>g>of</str<strong>on</strong>g> <str<strong>on</strong>g>of</str<strong>on</strong>g>fending is analyzed according to a Poiss<strong>on</strong><br />
regressi<strong>on</strong> model and several <str<strong>on</strong>g>of</str<strong>on</strong>g> its variants. When there is evidence <str<strong>on</strong>g>of</str<strong>on</strong>g> over-dispersi<strong>on</strong> –<br />
excess variati<strong>on</strong> not captured by the Poiss<strong>on</strong> distributi<strong>on</strong> – a negative binomial regressi<strong>on</strong><br />
model is examined. Because a substantial proporti<strong>on</strong> <str<strong>on</strong>g>of</str<strong>on</strong>g> our sample did not re<str<strong>on</strong>g>of</str<strong>on</strong>g>fend at<br />
75